This paper analyses the optimal stabilization policy when growth is driven by learning by doing. If benefits of learning by doing are not fully internalized, the optimal policy is to tax labor during expansions and to subsidize it during recessions. The long-term impact of this policy depends...